- Toyota Research Institute and Stanford Engineering have achieved the world’s first fully autonomous tandem drifting sequence.
- The development involved nearly seven years of collaboration to automate the motorsports maneuver of drifting.
- Autonomous drifting simulates challenging driving conditions, such as snow or ice, improving vehicle control and safety.
- The technology includes advanced AI techniques, such as neural network tire models, to emulate expert driving.
- Experiments were conducted using modified GR Supras at Thunderhill Raceway Park, with specialized modifications and AI algorithms.
- The vehicles used Nonlinear Model Predictive Control (NMPC) to continuously adjust steering, throttle, and braking.
- Autonomous drifting aims to enhance vehicle safety by providing advanced control in dynamic situations.
Main AI News:
In a groundbreaking development for automotive safety and artificial intelligence, Toyota Research Institute (TRI) and Stanford Engineering have achieved the world’s first fully autonomous tandem drifting sequence. This milestone in driving research highlights a significant advance in AI and its applications in vehicle control under dynamic conditions.
Over nearly seven years of collaborative effort, TRI and Stanford Engineering have perfected the art of autonomous drifting, a technique involving precise vehicle control after breaking traction by spinning the rear tires. This maneuver, integral to motorsports, is now being used to enhance vehicle safety, particularly in challenging conditions like snow or ice.
Avinash Balachandran, Vice President of TRI’s Human Interactive Driving division, emphasized the significance of this achievement. “Utilizing the latest tools in AI, we can now drift two cars in tandem autonomously. Mastering this complex maneuver opens new avenues for developing advanced safety systems in future automobiles,” Balachandran stated.
Chris Gerdes, Professor of Mechanical Engineering and Co-Director of the Center for Automotive Research at Stanford (CARS), highlighted the practical benefits of this technology. “The physics of drifting closely mimic what a car experiences on snow or ice. Our research has already led to new techniques for controlling automated vehicles in such conditions,” Gerdes noted.
The autonomous tandem drifting sequence involves two vehicles— a lead car and a chase car—navigating a course with inches of separation while operating at the edge of control. This process is facilitated by advanced AI techniques, including a neural network tire model that emulates the expertise of an experienced driver.
Gerdes added, “The AI developed for this project learns from every track session, adapting to variations in track conditions, such as changes in sunlight.”
Vehicle control during extreme conditions is crucial, as car crashes account for over 40,000 fatalities annually in the U.S. and approximately 1.35 million globally. The ability of autonomous systems to handle dynamic situations promises significant improvements in safety.
“When a vehicle begins to skid or slide, the driver must rely on their skills to avoid accidents. This new technology could act as an expert drifter, intervening precisely when needed to manage a loss of control,” Balachandran explained.
Gerdes concluded, “Achieving this feat shows what is possible in automotive safety. If we can master autonomous tandem drifting, the potential to enhance vehicle safety is immense.“
Technical Details
- The experiments were conducted at Thunderhill Raceway Park in Willows, California, using two modified GR Supras. TRI developed the algorithms for the lead car, while Stanford engineers worked on those for the chase car.
- TRI focused on stable control mechanisms for the lead car, while Stanford Engineering created AI models for the chase car to dynamically adapt and drift alongside without collisions.
- Modifications included suspension, engine, transmission, and safety systems by GReddy and Toyota Racing Development (TRD). The cars, built to Formula Drift specifications, allowed for data collection in a controlled environment.
- Both vehicles were equipped with computers and sensors to manage steering, throttle, and brakes, as well as a dedicated WiFi network for real-time communication about positions and trajectories.
- Autonomous tandem drifting was achieved through Nonlinear Model Predictive Control (NMPC), where each vehicle solves optimization problems up to 50 times per second to maintain drift and avoid collisions.
- AI training involved continual updates to the neural network using data from previous tests, improving performance with each track session.
Conclusion:
The achievement of fully autonomous tandem drifting by Toyota Research Institute and Stanford Engineering represents a significant advancement in automotive safety technology. This development not only demonstrates the potential for AI to handle complex driving maneuvers but also underscores the broader application of such technologies in enhancing vehicle safety under extreme conditions. As autonomous driving technologies continue to evolve, the ability to manage dynamic and challenging scenarios will likely become a critical factor in the development of safer and more reliable vehicles. This breakthrough could drive further innovation and investment in autonomous vehicle systems, potentially setting new standards for safety and performance in the automotive industry.
